Another possibility to reconcile?

Tuc at T-B-O-H.NET ml at t-b-o-h.net
Fri May 16 15:37:51 CEST 2008


Hi Ivan,

	We have many Chillispot systems, but as I mentioned before,
combined, at the time it was approximately 14.

	I was hoping that there would be some way to have checkrad do
that for us, but since DD-WRT runs Chillispot and not Coova, we don't have
access to a great utility Coova implemented which would be able to tell us
what sessions it has active at the time. 

	Since I wrote the email, I've learned more. It seems that the
DD-WRT units reboot daily as their version of "Garbage Collection". It
also seems that unlike Coova, the old Chilli did not send Accounting_On
when it booted, and Accounting_Off when it was shutting down. I have
to do some more log and file investigation, but I'm thinking that the
combination of those 2 issues is whats causing all the "dirty" sessions.

	Since all the units send output to syslog, I'm thinking about
having syslog output to a perl program that parses all the lines and if
it sees the beginning of a DD-WRT reboot, it will forge an Accounting_On
packet (If that is possible. I'm not sure if I'll run into the same issue
I did with trying to run a radtest from the radius server with the IP
and secret for a remote unit).

	I'll also see, but I'm sure it'll be pretty difficult, if I
can convince the DD-WRT people to port the Accounting_On/Off function,
and hopefully Coova itself!

	Thanks, Tuc
> 
> OK, you have 90 sessions open in radacct. How many users does Chillispot
> see as current? If there is a big difference than you are not getting
> all accounting Stop packets. You should run a script that removes stale
> entries (those open for longer than x hours) from radacct.
> 
> Ivan Kalik
> Kalik Informatika ISP
> 
> 
> Dana 16/5/2008, "Tuc at T-B-O-H.NET" <ml at t-b-o-h.net> pi¹e:
> 
> >Hi,
> >
> >	I'm looking to implement the Simultaneous-User Value in radcheck.
> >(FR 2.0.3) I'm having the issue that, for whatever reason (I'd blame the
> >network in a heartbeat, not FR at all), the accounting for a logged in user
> >never gets from a NULL acctstoptime to one filled in.
> >
> >	At the current time, radwho on the server shows approximately
> >22 active users. In reality I think it'd be more like 1/2 of that. A
> >"SELECT count( * ) FROM radacct WHERE acctstoptime IS NULL ;" shows 91
> >records.
> >
> >	Due to the version of the NAS we are running (DD-WRT with Chillispot),
> >we can't get checkrad to help true up the information.
> >
> >	Is there another way to help keep everything in sync, so we don't have
> >users who pay for a single ID, doing things like :
> >
> >lobnic14   00-13-02-25-8C-   shell S1   Thu 17:3  192.168.7 192.168.182.3
> >lobnic14   00-1B-77-11-F4-   shell S2   Thu 22:1  192.168.7 192.168.182.4
> >damrap6    00-0E-35-C0-16-   shell S1   Thu 22:1  192.168.5 192.168.182.5
> >damrap6    00-11-24-8F-27-   shell S3   Thu 20:2  192.168.5 192.168.182.10
> >damrap6    00-1B-77-06-2F-   shell S4   Thu 20:2  192.168.5 192.168.182.11
> >
> >		Thanks, Tuc
> >-
> >List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
> >
> >
> 
> -
> List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
> 
> 




More information about the Freeradius-Users mailing list